The article compares lexical doublets in English and Uzbek, focusing on how their meanings diverge and how their use is divided among styles and registers. In both languages, doublets arose mainly through contact with other languages: English with Scandinavian, French and Latin, and Uzbek with Persian, Arabic and, later, Russian and other European languages. The study distinguishes etymological doublets, which go back to the same source, from synonymic doublets, which are words of different origin with the same or very close meaning. It uses lexicological and lexicographic sources together with the author’s own contextual examples. Descriptive, comparativetypological, componential and contextual methods are applied. The results show three types of semantic variation, namely full divergence, partial divergence and stylistic differentiation. They also reveal a similar three-level functional model in both languages: native words serve everyday neutral speech, words borrowed in the earlier period serve literary and expressive speech, and learned borrowings serve formal, religious, scientific and official discourse. The study also shows that absolute doublets are unstable and tend either to differentiate or to drop out of use. The findings may be applied to teaching lexicology, to bilingual lexicography and to translation between English and Uzbek.